Meeting Banner
Abstract #2467

A Proximal Step Enables Fast and Accurate Single-Step QSM Reconstructions, Preventing Susceptibility Underestimation

Carlos Milovic1, Kristian Bredies2, Christian Langkammer3, and Karin Shmueli4
1University College London, London, United Kingdom, 2Institute of Mathematics and Scientific Computing, University of Graz, Graz, Austria, 3Department of Neurology, Medical University of Graz, Graz, Austria, 4Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom


Single-step quantitative susceptibility mapping (QSM) algorithms simplify the processing pipeline and promise to be more robust against background fields than traditional two-step methods but they often underestimate tissue susceptibilities. Here, we propose a highly efficient gradient descent Tikhonov-regularized proximal solver and a highly accurate ADMM TV-regularized proximal solver to improve the accuracy of two Laplacian-based single-step methods. Our solvers outperformed current single-step methods and showed in-vivo performance very similar to traditional two-step methods. This will simplify QSM processing pipelines, allowing further automation in future, although more research is needed to improve robustness against noise and boundary-conditions-related artifacts.

This abstract and the presentation materials are available to members only; a login is required.

Join Here